Optimized Clients for BOINC - Yes or No?

[petrusbroder]

You got it 100% AFAIK. 🙂

 

[BlackMountainCow]

Though, to a certain extent, the optimized PC will most of the time yield the highest result, which then will be deleted. Thus, the second highest - which actually should have been the highest - will count for the WU. So, you won't get the claimed credit but you'll get the highest "regular" one.

Does that make sense?

 

[Insidious]

No

Rosetta does not do redundancy checking. Only one result is returned and credited per work unit. What you claim is what you get.

Yes

For other BOINC projects that do redundancy checking

 

[BlackMountainCow]

Yes, yes, I was speaking for projects aside from Rosetta.

 

[SirUlli]

i dont like the Word cheating here

first:at BOINC all Results have to be validate and after this you get Credid

second: Make your own client software

everyone can use this, because goto first

and then at the SETI Community are people with very high Engagement, and only this and working together make this work

their are programmer, mathematics and physics and so on, only working together is one of the Reason the the the optimized Code is working so excellent.

some info

the main Seti Code is the Fourier transform, some Info is here

http://www.fftw.org/

and then the People at Berkely can only make Code that is running on every machine, so no SSE, SSE2, or somethink else is optimized.

but it make sense to optimize this, because more Workunits at a time, so i think that is also more Science,

only my personell Meaning, from an outstanding Member

Greetings from Germany NRW
Sir Ulli

 

[Rattledagger]

Originally posted by: networkman
If the optimized client does nothing to impede that, then I don't see any wisdom in not using an optimized client. For the possiblity of added credit to be realized, at least in E@H where redundancy checking IS enabled, there would have to be a SIGNIFICANT majority proportion of users also using optimized clients, otherwise as RD posted above, the higher credit results would be discarded anyway.

Somebody please correct me if I'm mistunderstanding. 🙂

Well, 99.99% of the time is used on the science-application and not the BOINC-client, so the only advantage would be higher benchmark.
But in SETI@Home, even there's 50% variations across my computers in their average claimed credits, they're within 5% of eachothers in average granted credit.

So, in projects appart of Rosetta@home, wouldn't expect any significant impact on average granted credit, and therefore doesn't see the reason for also downloading optimized BOINC-clients. Have been installing enough different BOINC-clients if shouldn't start to add optimized clients to the mix also...

Not to forget, some of the optimized BOINC-clients corresponds to "standard"-versions that has either failed alpha-testing, or haven't even been alpha-tested at all...

 

[SirUlli]

Originally posted by: Rattledagger

Originally posted by: networkman
If the optimized client does nothing to impede that, then I don't see any wisdom in not using an optimized client. For the possiblity of added credit to be realized, at least in E@H where redundancy checking IS enabled, there would have to be a SIGNIFICANT majority proportion of users also using optimized clients, otherwise as RD posted above, the higher credit results would be discarded anyway.

Somebody please correct me if I'm mistunderstanding. 🙂

Well, 99.99% of the time is used on the science-application and not the BOINC-client, so the only advantage would be higher benchmark.
But in SETI@Home, even there's 50% variations across my computers in their average claimed credits, they're within 5% of eachothers in average granted credit.

So, in projects appart of Rosetta@home, wouldn't expect any significant impact on average granted credit, and therefore doesn't see the reason for also downloading optimized BOINC-clients. Have been installing enough different BOINC-clients if shouldn't start to add optimized clients to the mix also...

Not to forget, some of the optimized BOINC-clients corresponds to "standard"-versions that has either failed alpha-testing, or haven't even been alpha-tested at all...

alpha testing, Beta testing...

the Results have to be validadet, is ths right yes...

You can do more Science in less time

and Berkeley says

http://boinc.berkeley.edu/anonymous_platform.php

make your own Client Software, or not....

the optimized client are buildt at the offical Berley Sources...

got to first

first:at BOINC all Results have to be validate and after this you get Credid

Sir Ulli

 

[Wolfsraider]

I am using them, and will continue to do so.
I do not believe this is cheating even in the smallest form.

Rosetta@Home has had no problems with it (as evidenced by Sid's post getting no negetive response from the admins) So as stated I will continue to use them.

Mike

 

[networkman]

I think I'm going to leave just Apophis with this optimized client for a couple weeks in case the situation should change. After there's a little more discussion on this topic, perhaps I'll optimize other clients as well.

 

[Kelemvor]

I don't think it's cheating since it's something that's available to everyone. It's all for fun anyway.

 

[Shuxclams]

I don't know how to get the optimized S@H to work... 🙁 I'd like to try it out, but get an error.

1/15/2006 7:03:08 PM|SETI@home|CreateProcess() failed - The system cannot find the file specified. (0x2)

SHUX

 

[BespinReactorShaft]

I'm sticking to the non-optimised client but am neutral on this issue i.e. do what you wanna do. I suppose that for power crunchers using optimised clients would be significant for overall team performance, but for me and my humble fleet of 2 rigs I'm mostly racing against myself. And if I wanted to reach some major milestone some time in the future I'd probably either a) get myself a brand new dual core rig :Q and/or b) get help from TABoo if possible. 😉

 

[Coquito]

Until something official is released, I'll stick to the old fashioned way of doing things.

 

[Malak]

Originally posted by: FrankyJunior
I don't think it's cheating since it's something that's available to everyone. It's all for fun anyway.

Cheats are available to everyone too.

 

[mondobyte]

Originally posted by: Malak

Originally posted by: FrankyJunior
I don't think it's cheating since it's something that's available to everyone. It's all for fun anyway.

Cheats are available to everyone too.

BOINC software is open source ... the clients may or may not be open source depending on the whims of the project (like PrimeGrid - refuses to release the source code -- and perhaps this violates the BOINC license too!)

In any case, as I understand it .... anyone can do anything to BOINC or the project client and as long as it gets "validated" correctly ... this is considered FAIR!

Now ... if folks don't know how to install the BOINC optimized clients ... I could provide the "transformed" MSI that could be downloaded provided that someone with enough bandwith wants to host it ...

You simply "repair" your current BOINC installation with the new MSI script and the optimized BOINC is installed.

This makes the playing field "level" for the non-techies who wouldn't know what to do with a binary in Windows. Simply use the "optimized" MSI to "install" (new) or "repair" (existing) BOINC client.

I'd have to think about how to make this "foolproof" for Linux users but I am sure it can be done easily enough. The problem is there are so many flavors of Linux out there ... a single technique might not work universally.

mondo

 

[Fardringle]

Is there a separate optimized client for each type of CPU (i.e. one for AMD, another for Intel, or even broken down by cores)? Or is it just one 'optimized' client that actually recognizes the abilities of the CPU better and gives it a better benchmark score?

 

[GLeeM]

Do I understand right that the new soon coming out Optimized Seti will need this BOINC optimization?
Then all Seti crunchers will have it, so how can it be cheating.

What is preventing someone from modifying the open source code to multiply the benchmark by 2 or 10 or ?

 

[mondobyte]

Originally posted by: GLeeM
Do I understand right that the new soon coming out Optimized Seti will need this BOINC optimization?
Then all Seti crunchers will have it, so how can it be cheating.

What is preventing someone from modifying the open source code to multiply the benchmark by 2 or 10 or ?

You are confusing the imminent release of "Enhanced SETI" with "optimized SETI Clients" ... and yes ... there will be multiple "flavors" of Optimized Enhanced SETI Clients ...

Nothing ... except validation of the returned work by SETI. If it validates ... then it is up to the project how it interprets or uses or doesn't use the benchmark.

As I understand it, SETI Enhanced is moving towards the direction of accumulating flops within the SETI client making BOINC benchmarks a floccinaucinihilipilification!

mondo

 

[petrusbroder]

@ Fardingdale
Yes there are separate clients for diffrent processors, but not based on brand but on the functions implemented in them (i.e. MMX, SSE, SSE2 etc.). You have to picke the right file for your processor - and to find out what is implemented you cna use a program such as CPU-Id or CPU-z.

@GLeeM:
Seti is coming out with an enhabced application, which does a lot mor crunching with the WUs - i.e. no optimized client at all. You could look at the client as a caretaker, who takes care of benchmarking, input/output etc., while the application does the crunching. And there is nothing preventing anybody with knowledge to write their own client and to multiply the benchmeark (or even the claimed credits) by a factor of "anything". I have seen such clients ... it doesn not help inthose projects (seti, einstein predictor etc) where several WU's results and credits are compared. Thus such effort is more or less useless. AFAIK only Rosetta grants all the credit claimed ...

 

[Wiz]

I voted: YES, NO, NO

As always, we are all free to use & do whatever we want with our resources.

Use Optimized Clients or don't - it's up to you.

For the same reasoning that I stuck to Seti Classic 3.03 when 3.08 came out, I am and will continue to use optimized clients.

There are two things going on at once here - try to wrap your head around two things at the same time:

1. the science - which does not suffer due to using the optimized client
2. the competition - which does suffer if those ahead of us (and behind us) use them and we do not

In Seti1 cheating meant that you returned bogus results or the same results over and over. That was bad for the science (and your team). That has nothing whatsoever to do with these optimized clients, that is not what is happening. Instead you are returning good & accurate results and you are getting as much credit as your machines can get for doing it.

What would be bad for the science is deciding to simply turn off a bunch of computers because some "individuals" decided to go on some kind of crusade and started calling you a "cheater" and just made you sick of spending time around great places like the Anandtech DC forum.

The people in charge of Boinc & Rosetta either have no comment or are not against the use of these clients so it is up to personal choice.

The NO vote on number 3 is because we don't need, and will never reach an agreement either way. If you can't live with our diversity - don't let the door hit your ass on the way out. That's my point of view, and I'm sticking with it.

 

[mondobyte]

My sentiments exactly Wiz ... well said.

My offer to assist those who wish to run and optimized BOINC still stands ...

 

[Kelemvor]

I have my BOINCs installed as a Service. Anyone know if I have to actually stop the service or do anything funky before installing the optimized clients?

EDIT: N/M. Looks like I have to manually stop the service, then extract the stuff, then start the service up again.

Will post the number differences when I'm done with all the PCs.

 

[BadThad]

Originally posted by: Fardringle
Is there a separate optimized client for each type of CPU (i.e. one for AMD, another for Intel, or even broken down by cores)? Or is it just one 'optimized' client that actually recognizes the abilities of the CPU better and gives it a better benchmark score?

Read it all here

 

[Malak]

The science may not suffer, but the competition does. I'm sorry, but knowing that one person doing less science can get more points just because they use a special client... that doesn't sound like cheating to any of you? It doesn't matter if it is open source, it doesn't matter if anyone can get it... it's still cheating. In any and every competition in the world that is cheating. Even if the administration doesn't care, I will still think it's cheating. Obviously the administration doesn't care about the competition itself, only the science.

 

[GimpyOne]

Originally posted by: Malak
The science may not suffer, but the competition does. I'm sorry, but knowing that one person doing less science can get more points just because they use a special client... that doesn't sound like cheating to any of you? It doesn't matter if it is open source, it doesn't matter if anyone can get it... it's still cheating. In any and every competition in the world that is cheating. Even if the administration doesn't care, I will still think it's cheating. Obviously the administration doesn't care about the competition itself, only the science.

But are the swimmers at the oympics wearing the full body slick suits cheating because not everyone from every country can afford to buy them? Maybe yes, maybe no.(rules say no and some of these people are beaten by those wearing standard speedos; do they really help in the end?) So becuase someone can't/won't install the optimized client, is using one cheating? As is said here it is personal opinion; which will soon become a non-existant issue as flops counting is enabled.(then something else will become an issue!) People have asked the "rules committee", ie...the Devs...and they have not voiced issues with the optimized clients.

I think as long as the science isn't damaged as it was with the Classic cheaters, then it really isn't a big deal which someone uses. I personally am not becuase I'm too lazy to do it! These races are for fun and to contribute to what you feel is a good cause, not money, power, or even ginormous e-genetalia!

Well, just my 2cents worth.